OFA Is a Tool, Not a Verdict

Abstract

The Orthopedic Foundation for Animals (OFA) hip evaluation is one of the best-known canine health-screening tools in North America. It is also frequently misunderstood. OFA hip screening is a phenotypic radiographic evaluation: it describes the conformation and radiographic condition of a dog’s hips at the time of evaluation, using a standardized hip-extended view reviewed by veterinary radiologists. That information is meaningful and heritable, but it is not equivalent to a DNA-based genetic test and it does not directly measure the dog’s additive genetic breeding value. Canine hip dysplasia (CHD) is a complex, polygenic developmental disorder whose expression is influenced by joint laxity, breed-specific anatomy and biomechanics, growth, body weight, nutrition, exercise, and other environmental effects. [1-6]

Long-term data show that radiographic selection can improve hip conformation at the population level, but the magnitude of improvement varies among breeds, raw registry statistics are vulnerable to voluntary-submission bias, and research consistently suggests that selection based on family information and estimated breeding values can outperform selection based on an individual score alone. [7-11] Population genetics adds another essential constraint: maximizing selection pressure on one health trait can reduce effective population size and genetic diversity, especially if carriers of well-managed recessive variants or large fractions of phenotypically normal dogs are removed from reproduction unnecessarily. [12-16] The appropriate scientific conclusion is therefore neither “OFA is everything” nor “OFA does not work.” OFA is valuable data. Its power depends on how intelligently that data is interpreted and incorporated into a broader, population-level breeding strategy.

A useful distinction

Phenotype: what can be observed or measured in the dog.

Genotype: the dog’s DNA sequence or genotype at a particular locus.

Breeding value: the portion of an animal’s genetic merit expected, on average, to be transmitted to offspring for a trait. It is estimated - not read directly from an X-ray.

1. What OFA Actually Measures

OFA hip screening is a radiographic phenotype assessment. For dogs 24 months of age or older, OFA evaluates a standardized ventrodorsal hip-extended radiograph. Three board-certified veterinary radiologists independently assess the image, and the consensus result is classified as Excellent, Good, Fair, Borderline, Mild, Moderate, or Severe. Excellent, Good, and Fair are considered within normal limits. OFA explicitly describes these categories as phenotypic hip conformations. [1,2]

That distinction is not semantic. The image contains information about the dog’s developed joint anatomy: femoral-head seating, acetabular coverage, congruity, subluxation, remodeling, and secondary osteoarthritic change. It is therefore a clinically and biologically meaningful phenotype. But a radiograph is not a readout of all the alleles that contributed to that phenotype, and it cannot separate every genetic contribution from every developmental, biomechanical, or environmental contribution.

OFA itself emphasizes that breed, sex, age, positioning, and radiographic quality matter to interpretation. Poor positioning can make one socket appear shallower and another deeper, which is why technically inadequate studies are rejected and repeat imaging may be requested. OFA also notes that periods of prolonged inactivity can temporarily affect apparent subluxation in some animals and recommends imaging dogs in good health and muscular condition. [2,17]

Genetic testing is not the same thing as health screening

The common shorthand that “health testing only tells you the dog’s status that day” therefore needs refinement. That description is reasonably intuitive for a clinical eye or cardiac examination, where disease may emerge later. A mature hip radiograph is more than a momentary clinical snapshot: it records structural development that has occurred over time. Yet it remains a phenotype, not a direct assay of the genes responsible for that structure. [1,2]

2. Hip Dysplasia Is a Complex Quantitative Trait

Canine hip dysplasia is not a classic single-gene disorder. It is a developmental, complex trait characterized by varying combinations of joint laxity, subluxation, incongruity, abnormal loading, and eventual osteoarthritis. Multiple genetic loci contribute to susceptibility, and studies have identified quantitative-trait loci and susceptibility markers rather than a single universal “hip dysplasia gene.” [3,4,18]

This matters because complex traits behave statistically. A dog can inherit a relatively high or low liability from many alleles of small or moderate effect, and the phenotype that becomes visible is then influenced by the developmental environment. In such traits, a normal individual can still carry above-average genetic liability, while an abnormal individual can have a phenotype that is worse than its underlying transmissible genetic liability would suggest.

Heritability: useful, but commonly misunderstood

Heritability (h²) is the proportion of phenotypic variance in a particular population and environment that is attributable to additive genetic variance. It is not the percentage of an individual dog’s hips that is “genetic.” It is also not fixed across all breeds or all populations.

In Bernese Mountain Dogs, published estimates illustrate that point. A German dataset of 5,534 Bernese Mountain Dogs estimated CHD heritability at 0.26, while a Swedish analysis of 8,221 Bernese Mountain Dogs reported hip- and elbow-dysplasia heritabilities in the 0.34-0.42 range. Both studies concluded that breeding values can improve selection efficiency beyond the individual phenotype. [9,10]

Moderate heritability is exactly the scenario in which phenotypic screening is useful but imperfect. There is enough genetic signal for selection to work, yet enough residual variation that the individual score is not a pure genetic measurement.

The breeder’s equation - conceptually

In quantitative genetics, response to selection is often summarized as R = h²S: the genetic response (R) depends on heritability (h²) and the selection differential (S).

The practical implication is not that “harder selection is always better.” Stronger selection can increase short-term response for one trait while simultaneously reducing diversity, effective population size, or progress on other traits. Multi-trait population management matters.

3. Environment Changes the Expression of Hip Risk

The statement “hip dysplasia is genetic” is true but incomplete. OFA describes CHD as inherited and also notes that environmental factors can affect severity. Veterinary reviews similarly describe a consensus that genetic predisposition interacts with environmental and biomechanical factors. [1,5,6]

Growth rate, caloric intake, and body condition

Classic controlled studies demonstrated that rapid weight gain produced by higher caloric intake increased the frequency, earlier appearance, and severity of hip dysplasia in genetically susceptible dogs. In a paired Labrador Retriever study, dogs fed 25% less food than their matched littermates showed less subluxation and less radiographic hip dysplasia at two years. A later longitudinal analysis of the same broad research colony found that lifelong diet restriction reduced radiographic hip osteoarthritis. [19-21]

These findings should not be interpreted as “food causes hip dysplasia.” A more accurate statement is that nutrition and body mass can modify the expression and clinical consequences of inherited susceptibility. Excess calcium during growth has also been identified as a developmental risk factor, particularly in large and giant breeds. [5,19]

Exercise and the developmental environment

Exercise effects are more nuanced than internet rules often imply. In a prospective Norwegian cohort of 501 Newfoundlands, Labrador Retrievers, Leonbergers, and Irish Wolfhounds, stair use from birth to three months was associated with increased radiographic CHD risk, while off-leash exercise over soft, moderately rough terrain during that period was associated with reduced risk. The study also found strong litter clustering, underscoring the simultaneous importance of inherited factors. [22]

This is evidence for developmental modulation, not a universal prohibition on all stairs or a guarantee that a particular exercise regimen will prevent CHD. Breed, intensity, age, surface, body condition, and underlying susceptibility all matter.

Muscle, loading, and biomechanics

Hip phenotype emerges in a mechanical system. Joint laxity changes loading; abnormal loading changes cartilage and bone; muscle mass can stabilize a lax joint; and body weight changes the forces transmitted through the joint. This is why the same radiographic category does not produce identical clinical disease in every dog and why some dogs with marked arthritic change remain highly functional while others with less dramatic radiographic change are painful. OFA itself notes this imperfect relationship between radiographic severity and clinical signs. [1]

Genes establish liability. Development and biomechanics help determine how that liability is expressed.

4. Breed Architecture Matters: Why Greyhounds Are So Informative

Perhaps the clearest way to understand breed-level orthopedic risk is to compare populations with profoundly different hip phenotypes. Performance Greyhounds and Borzois have historically shown an extremely low prevalence of CHD and unusually tight hip joints. In foundational distraction-radiography work, these breeds consistently had low distraction indices, while breeds with high CHD prevalence had greater passive laxity. [23]

Researchers exploited this contrast scientifically. Cornell investigators created Labrador Retriever × Greyhound pedigrees by mating dysplastic Labradors to trait-free Greyhounds. In those crosses, hip laxity and other dysplasia-related traits changed progressively with the proportion of Labrador versus Greyhound ancestry, supporting a strong inherited component to breed-level hip architecture. [24,25]

The Greyhound example should not be reduced to “Greyhounds have good hips because they are shaped differently.” The more precise conclusion is that breeds differ in inherited anatomy, joint laxity, growth pattern, musculature, locomotor mechanics, and the underlying frequencies of risk and protective alleles. Visible conformation is part of that system, but it is not the entire causal explanation.

This is also why a universal numerical threshold does not always mean exactly the same thing biologically across breeds. In a study of 4,349 dogs from four large breeds - including Bernese Mountain Dogs - PennHIP distraction index was the strongest measured risk factor for hip osteoarthritis, but the relationship between laxity and osteoarthritis was breed-specific. [26]

The population lesson from Greyhounds

Different breeds do not begin with the same baseline genetic liability for CHD.

A screening score should therefore be interpreted against breed and family context, not as though every canine population has the same distribution of anatomy, laxity, and disease risk.

Selection changes populations over generations; it does not turn a complex trait into a single-gene disorder.

5. Dog A, Dog B, and Dog C: Why Context Changes the Meaning of a Score

A useful thought experiment is to imagine three dogs. The point is not that a breeder should override a radiologist. The point is that an individual phenotype is only one observation from which we are trying to infer transmissible genetic risk.

Dog C: what trauma changes - and what it does not

OFA notes that hip pathology can arise not only from abnormal joint development but also from cartilage damage following a traumatic fracture. [1] That makes an important conceptual point: an abnormal radiograph can describe a real phenotype without proving that every component of that phenotype arose from inherited developmental dysplasia. However, “there was an injury” should never be used as a blanket explanation for an unfavorable screening result. A defensible interpretation would look for objective evidence: veterinary records documenting the injury, its location and timing, whether the abnormality is unilateral or bilateral, the pattern of remodeling, prior imaging if available, the unaffected hip, orthopedic or radiology specialist opinion, and - critically - the orthopedic history of relatives and offspring.

In other words, Dog C does not automatically become a good breeding candidate. But neither should population genetics pretend that every acquired lesion is an inherited liability allele. The purpose of interpretation is to separate, as far as the evidence permits, transmissible genetic signal from non-transmissible noise.

Phenotype is evidence about genotype. It is not identical to genotype.

6. What Decades of OFA Screening Have - and Have Not - Accomplished

The most scientifically defensible position is not that OFA screening has failed. Large datasets show that selection using radiographic hip phenotypes has produced real improvement. The more important question is how much improvement, in which breeds, under which breeding structures, and whether better genetic tools could accelerate progress.

Evidence that screening works

A 2017 analysis of more than one million OFA hip records and 275,000 elbow records across 60 breeds found overall improvement in hip and elbow conformation and reductions in estimated breeding values for disease liability. The magnitude of response varied substantially by breed, and the authors emphasized that selection strategies should be breed-specific. [7]

A separate analysis of 74 breeds and more than 1.2 million dogs found clear genetic improvement after adjustment for reporting bias, age, and pedigree - but described the overall advance as modest. For hip dysplasia, the total estimated genetic improvement over roughly four decades was 16.4% of a phenotypic standard deviation. Raw phenotypic scores showed much less obvious change until statistical models addressed voluntary reporting and other influences. [8]

Bernese Mountain Dog data also argue against an “OFA never changes anything” narrative. Swedish Bernese data showed favorable genetic trends, and a Swiss retrospective study reported a decline in screened Bernese CHD prevalence from 21% in 1995-1999 to 12% in 2010-2016. Importantly, the Swiss authors warned that only about 23% of Bernese Mountain Dogs were radiographed, so those screened-dog percentages could not be treated as true population prevalence. [9,27]

Why registry percentages are not the same as population prevalence

Voluntary databases are vulnerable to selection bias. Owners may be more likely to submit favorable films, veterinarians may recognize obvious dysplasia before formal submission, and abnormal results have historically been less likely to enter public-facing datasets. The 2013 OFA analysis explicitly described severe reporting bias, especially in earlier decades. OFA-affiliated authors have also acknowledged prescreening and non-submission as a source of downward bias in apparent dysplasia frequency. [8,11]

Therefore, an OFA breed statistic is best understood as a statistic about the evaluated and submitted population, not a census of every dog in the breed. This does not make the database useless; it means the denominator must be understood.

The accurate conclusion

Radiographic selection has improved hip phenotypes in many canine populations.

Progress has been uneven and often slower than a simple “pass/fail” model would lead the public to expect.

Voluntary-submission bias means raw registry percentages should not automatically be interpreted as true breed prevalence.

The evidence favors better use of the data - especially relatives, progeny, EBVs, and complementary phenotypes - rather than abandonment of screening.

7. OFA and PennHIP Measure Related but Different Information

OFA and PennHIP are often framed as competitors, but scientifically they emphasize different components of hip phenotype. OFA uses the hip-extended view to assess overall conformation, congruity, subluxation, and secondary degenerative change. PennHIP includes compression and distraction views and quantifies passive laxity using the distraction index (DI). [2,28]

The distinction matters because the standard hip-extended position can reduce apparent passive laxity. In a study of 439 adult dogs that underwent both systems, many dogs rated Excellent, Good, or Fair by OFA still had distraction indices of 0.30 or greater; overall, 80% of dogs with OFA-normal ratings in that sample had DI values at or above 0.30. The authors concluded that the hip-extended OFA score can miss clinically relevant passive laxity detected by distraction radiography. [29]

That finding should also be interpreted carefully. DI is a measure of laxity and risk, not a guarantee of future pain or clinical disease. Breed-specific relationships between laxity and osteoarthritis exist, and clinical outcome depends on more than one measurement. [11,26]

A 2020 study of 615 purpose-bred detection dogs is especially informative for selection. Hip quality did not improve during a period in which OFA scores alone were used for breeding selection; after a PennHIP DI threshold was added to the program, hip quality improved significantly. This is strong evidence that complementary phenotypes can increase selection efficiency in a structured population, although it should not be generalized as proof that every breeding population must use the identical threshold or protocol. [30]

8. From Individual Scores to Estimated Breeding Values

If the true breeding objective is to reduce inherited liability, the statistically relevant target is not the individual phenotype itself. It is the animal’s breeding value: the additive genetic merit expected to be transmitted, on average, to offspring. Estimated breeding values (EBVs) use statistical models to combine the animal’s own phenotype with information from genetically related animals, while accounting for pedigree relationships and, where models allow, fixed environmental effects. A dog with many evaluated siblings, parents, grandparents, and offspring can therefore be estimated more accurately than a dog with the same individual score but no family information. [8-11,31]

This is not a fringe criticism of OFA. OFA’s own basic breeding recommendations explicitly advise breeders to consider normal parents and grandparents, the proportion of normal siblings, and the percentage of normal progeny, concluding that breeding goals depend on assessing predictive breeding value. [32]

Why Dog A and Dog B are not genetically equivalent

Suppose Dog A and Dog B are both OFA Good. If Dog A comes from a family with consistently favorable hips and Dog B comes from a family with repeated dysplasia, the two dogs have the same categorical phenotype but not necessarily the same expected genetic merit. Family information changes the posterior probability of what each individual is likely to transmit.

The logic works in the other direction as well. A single unexpectedly poor phenotype in an otherwise strong family should be taken seriously, but it can be weighted differently from a poor phenotype that sits inside a repeated familial pattern. EBVs formalize that logic rather than relying only on breeder intuition.

Why EBVs can outperform individual selection

Reviews of CHD control have repeatedly recommended EBV-based selection for quantitative traits. Swedish Bernese and Rottweiler data predicted faster progress from breeding values than from phenotype alone, and the large U.S. OFA analyses reached the same general conclusion. [8-10,31]

In practical terms, the objective should be to improve the expected genetic distribution of the next generation, not to collect the prettiest certificate for each parent.

9. The Missing Public Conversation: Population Genetics

Public discussions of “responsible breeding” are often framed as an individual checklist: hips, elbows, eyes, heart, DNA panel. Those tests matter. But breeding is not fundamentally an individual-dog exercise. It is the deliberate alteration of allele frequencies and trait distributions across a population over generations.

That shift in perspective changes almost every question. Instead of asking only whether a dog is ideal, population genetics asks what happens if thousands of breeders apply the same rule, generation after generation.

Selection has opportunity costs

A breeding population has finite size. Every exclusion removes not only the targeted risk allele or phenotype but also the rest of that animal’s genome. If selection against one trait is too intense, too rapid, or too concentrated around a small number of favored animals, effective population size can fall and inbreeding can rise. [12-16]

This is why the seemingly virtuous rule “only breed the absolute best hip score” can become genetically simplistic. OFA-affiliated authors have cautioned that restricting reproduction to only the highest hip categories would unnecessarily narrow breed gene pools; their recommended alternative is sustained selection for normal hips informed by parents, grandparents, siblings, progeny, and EBVs. [11]

Effective population size is not the same as head count

A breed can contain many thousands of registered dogs yet have a much smaller effective population size if a limited number of animals contribute disproportionately to the next generation. Pedigree analyses of purebred dogs have documented low effective population sizes in multiple breeds and major losses of genetic diversity associated with closed populations and unequal reproductive contribution. [15,33]

This is the population-genetic reason the “popular sire” effect matters. A male can be healthy, well tested, and genetically valuable - and still become harmful to population resilience if he contributes an excessive fraction of the next generation. Rare deleterious alleles carried by any dog can become common before their consequences are recognized. [14,15]

DNA carriers illustrate the same principle even more clearly

For a well-validated autosomal recessive disorder, a carrier bred to a clear dog does not produce genetically affected offspring. If the disease allele is common, eliminating every carrier in one generation may remove a large amount of unrelated genetic diversity for no immediate welfare benefit. Veterinary genetic counseling literature therefore recommends, in appropriate circumstances, breeding high-quality carriers to clear dogs and replacing them over time with clear descendants rather than abruptly discarding entire lines. [12,13]

Simulation studies show that the genetic-diversity consequences of removing carriers depend on allele frequency and breed structure, while limiting excessive use of individual sires can help preserve diversity. [14] DNA testing is therefore most powerful when it enables safer matings - not when every positive result is automatically converted into permanent exclusion.

The same logic applies to complex traits, but with more uncertainty

There is no single “carrier” state for CHD. Instead, every dog carries a distribution of risk and protective alleles. This makes population management harder: breeders must apply directional pressure without pretending that a categorical radiographic grade is a complete genetic diagnosis. The best available solution is richer data - broad screening, family information, longitudinal outcome data, EBVs, and eventually genomic prediction validated within the relevant population.

10. A Dog Is Not a Pair of Hips

Population selection is multi-trait by necessity. Orthopedic health matters, but so do cardiac health, eyes, neurologic disease, cancer risk, immune function, fertility, temperament, structure, longevity, and genetic diversity. Selection pressure applied to one dimension can alter the distribution of others, especially when the breeding pool is small.

Bernese Mountain Dog research offers a concrete example. Multivariate analysis has found genetic correlations among hip dysplasia, elbow dysplasia, and osteochondrosis. Those correlations are not necessarily large or always favorable, which means selection decisions can have correlated responses in other traits. [10] This is another reason multi-trait breeding values are more sophisticated than a collection of independent pass/fail rules.

The general public often imagines that adding more exclusions must monotonically increase health. Population genetics says otherwise. Better testing increases information. Better decisions require deciding how strongly to act on each piece of information while preserving the population’s ability to improve tomorrow.

11. A More Scientifically Defensible Breeding Framework

A modern health program can take OFA seriously without treating it as the end-all, be-all. The following framework is consistent with the evidence reviewed above:

1. Screen broadly and consistently. Phenotypic data become more useful when a high proportion of the population is evaluated, including pets and animals not ultimately bred. Broad screening reduces the bias created when only likely “good” dogs are submitted.

2. Keep abnormal results visible whenever possible. Open reporting improves population-level inference. A database full of only successful certificates is less useful for estimating family risk than a database that captures the full phenotype distribution.

3. Interpret the individual in family context. Parents, grandparents, full and half siblings, and prior offspring can materially change confidence in the individual score. OFA itself recommends this approach.

4. Distinguish inherited developmental disease from acquired pathology. A documented traumatic lesion, infection, surgery, or other acquired process should be medically characterized rather than automatically treated as evidence of inherited CHD - while avoiding unsupported “injury” excuses for developmental dysplasia.

5. Use complementary measurements when they add information. OFA conformation, PennHIP laxity, clinical history, and specialist evaluation measure overlapping but non-identical aspects of orthopedic risk.

6. Prefer EBVs or comparable statistical tools when the data infrastructure exists. The goal is prediction of genetic merit, not celebration of one phenotype.

7. Manage recessive DNA variants rather than reflexively purging carriers. For validated simple recessives, safe carrier-to-clear matings can preserve diversity while the allele frequency is reduced over generations.

8. Limit popular-sire effects. The best-tested dog in the population can still be overused. Genetic contribution should be distributed across multiple valuable lines.

9. Select across the whole dog. Orthopedic health should be balanced against other health, temperament, functional, structural, and diversity goals.

10. Audit population trends, not just individual compliance. A program should ask whether the incidence and breeding values of important disorders are actually improving over generations, while tracking diversity and unintended correlated changes.

What the evidence does not justify

“OFA is useless.” False. Large datasets demonstrate genetic and phenotypic improvement from radiographic selection. “An OFA Good dog has good hip genetics.” Too strong. It has a favorable individual phenotype; genetic merit is better estimated with relatives and progeny.

“A dog with an abnormal radiograph should always be removed from breeding, regardless of cause.” Too categorical. Acquired trauma can alter hip pathology; the cause and the population context matter. This requires veterinary documentation, not assumption.

“Environment causes hip dysplasia.” Misleading. Genetic liability is central, while environment and biomechanics modify expression and severity.

“DNA clear means genetically healthy.” False. A DNA panel covers specific known variants. It cannot certify the rest of the genome, complex disease risk, or future acquired health.

“More exclusions always make a population healthier.” False. Selection can reduce disease risk while also reducing diversity or concentrating other harmful alleles if population structure is ignored.

12. Conclusion: Data Are the Beginning of a Breeding Decision, Not the End

OFA deserves neither blind reverence nor casual dismissal. It is a standardized, specialist-interpreted source of phenotypic information that has contributed to measurable improvement in canine hip conformation. The scientific limitation is not that the data are meaningless. The limitation is that phenotype is only one component of genetic inference.

A mature hip radiograph records the result of an entire developmental process. That process reflects inherited liability, breed-specific anatomy and biomechanics, growth, body condition, exercise, and sometimes acquired injury. The same OFA grade can therefore carry different genetic implications in different families, and an abnormal phenotype can have different causes.

Population genetics provides the missing framework. The goal of breeding is not to identify flawless individuals. It is to shift the distribution of health and function in the next generation while preserving enough genetic diversity to continue improving future generations. That requires screening, but it also requires family data, progeny data, quantitative genetic tools, restraint in the use of popular sires, intelligent management of recessive variants, and attention to the whole dog.

The most useful way to think about OFA is therefore simple:

OFA gives us data. Population genetics tells us how to use it.

Editorial / scientific caution

This article is an educational review of breeding science, not a protocol for deciding whether any individual dog should be bred. Dogs with abnormal radiographs, prior orthopedic trauma, pain, lameness, or ambiguous imaging should be evaluated by the attending veterinarian and, when appropriate, a board-certified veterinary radiologist or orthopedic surgeon.

No single study establishes a universal breeding threshold for every breed or crossbred population. The strongest programs combine objective testing with transparent longitudinal data and population-specific outcome monitoring.

References

1. Orthopedic Foundation for Animals. “Hip Dysplasia.” Official disease overview and grade descriptions. Accessed October 2026. Link
2. Orthopedic Foundation for Animals. “Hip Screening Procedures.” Official radiographic protocol and evaluation process. Accessed October 2026. Link
3. King MD. Etiopathogenesis of Canine Hip Dysplasia, Prevalence, and Genetics. Vet Clin North Am Small Anim Pract. 2017;47(4):753-767. doi:10.1016/j.cvsm.2017.03.001. PMID 28460694. Link
4. Zhou Z, et al. Differential genetic regulation of canine hip dysplasia and osteoarthritis. PLoS One. 2010;5:e13219. PMID 20949002. Link
5. Fries CL, Remedios AM. The pathogenesis and diagnosis of canine hip dysplasia: a review. Can Vet J. 1995;36(8):494-502. PMID 7585436; PMCID PMC1687006. Link
6. Schachner ER, Lopez MJ. Diagnosis, prevention, and management of canine hip dysplasia: a review. Vet Med (Auckl). 2015. PMID 30101105. Link
7. Oberbauer AM, Keller GG, Famula TR. Long-term genetic selection reduced prevalence of hip and elbow dysplasia in 60 dog breeds. PLoS One. 2017;12(2):e0172918. doi:10.1371/journal.pone.0172918. Link
8. Hou Y, Wang Y, Lu X, et al. Monitoring Hip and Elbow Dysplasia achieved modest genetic improvement of 74 dog breeds over 40 years in USA. PLoS One. 2013;8(10):e76390. doi:10.1371/journal.pone.0076390. Link
9. Malm S, Fikse WF, Danell B, Strandberg E. Genetic variation and genetic trends in hip and elbow dysplasia in Swedish Rottweiler and Bernese Mountain Dog. J Anim Breed Genet. 2008;125(6):403-412. doi:10.1111/j.1439-0388.2008.00725.x. Link
10. Hartmann P, Stock KF, Distl O. Multivariate genetic analysis of canine hip and elbow dysplasia as well as humeral osteochondrosis in the Bernese Mountain Dog. Berl Munch Tierarztl Wochenschr. 2010;123:488-495. PMID 21141279. Link
11. Keller GG, Dziuk E, Bell JS. How the Orthopedic Foundation for Animals is tackling inherited disorders in the USA: using hip and elbow dysplasia as examples. Vet J. 2011;189(2):197-202. Link
12. Bell JS. Researcher responsibilities and genetic counseling for pure-bred dog populations. Vet J. 2011;189(2):234-235. doi:10.1016/j.tvjl.2011.06.025. Link
13. Wade CM. Inbreeding and genetic diversity in dogs: results from DNA analysis. Vet J. 2011;189(2):183-188. doi:10.1016/j.tvjl.2011.06.017. Link
14. Leroy G, Rognon X, Varlet A, Joffrin C, Verrier E. Assessing the impact of breeding strategies on inherited disorders and genetic diversity in dogs. Vet J. 2012;194(3):343-348. PMID 22819181. Link
15. Leroy G, Baumung R. Mating practices and the dissemination of genetic disorders in domestic animals, based on the example of dog breeding. Anim Genet. 2010;42:66-74. PMID 20528847. Link
16. Farrell LL, Schoenebeck JJ, Wiener P, Clements DN, Summers KM. The challenges of pedigree dog health: approaches to combating inherited disease. Canine Genet Epidemiol. 2015;2:3. PMID 26401331. Link
17. Orthopedic Foundation for Animals. The Use of Health Databases and Selective Breeding: A Guide for Dog and Cat Breeders and Owners. See discussion of positioning, inactivity, and database interpretation. Link
18. Zhu L, et al. Quantitative trait loci for hip dysplasia in a cross-breed canine pedigree. Mamm Genome. 2005. PMID 16245029. Link
19. Richardson DC. The role of nutrition in canine hip dysplasia. Vet Clin North Am Small Anim Pract. 1992;22(3):529-540. PMID 1604770. Link
20. Kasström H. Nutrition, weight gain and development of hip dysplasia: an experimental investigation in growing dogs with special reference to feeding intensity. Acta Radiol Suppl. 1975. PMID 1066031. Link
21. Kealy RD, et al. Effects of limited food consumption on the incidence of hip dysplasia in growing dogs. J Am Vet Med Assoc. 1992;201(6):857-863. PMID 1399793; and Smith GK, et al. Lifelong diet restriction and radiographic evidence of hip joint osteoarthritis in dogs. JAVMA. 2006;229(5):690-693. PMID 16948575. Link
22. Krontveit RI, Nødtvedt A, Sævik BK, et al. Housing- and exercise-related risk factors associated with the development of hip dysplasia in a prospective cohort of four large dog breeds. Am J Vet Res. 2012;73(6):838-846. PMID 22620698. Link
23. Smith GK. Advances in diagnosing canine hip dysplasia. J Am Vet Med Assoc. 1997. PMID 9154196. Link
24. Todhunter RJ, et al. An outcrossed canine pedigree for linkage analysis of hip dysplasia. J Hered. 1999;90(1):83-92. PMID 9987910. Link
25. Todhunter RJ, et al. Quantitative genetics of traits associated with hip dysplasia in a canine pedigree constructed by mating dysplastic Labrador Retrievers with unaffected Greyhounds. Am J Vet Res. 2002. PMID 12118665. Link
26. Runge JJ, Kelly SP, Gregor TP, Kotwal S, Smith GK. Distraction index as a risk factor for osteoarthritis associated with hip dysplasia in four large dog breeds. J Small Anim Pract. 2010;51(5):264-269. doi:10.1111/j.1748-5827.2010.00937.x. Link
27. Ohlerth S, Geiser B, Flückiger M, Geissbühler U. Prevalence of Canine Hip Dysplasia in Switzerland Between 1995 and 2016 - a retrospective study in five common large breeds. Front Vet Sci. 2019. PMID 31709271. Link
28. Smith GK, Biery DN, Gregor TP. New concepts of coxofemoral joint stability and development of a clinical stress-radiographic method for quantitating hip joint laxity in the dog. JAVMA. 1990;196:59-70. PMID 2295555. Link
29. Powers MY, et al. Evaluation of the relationship between Orthopedic Foundation for Animals hip joint scores and PennHIP distraction index values in dogs. JAVMA. 2010;237:532-541. PMID 20807130. Link
30. Lofton SC, et al. Effectiveness of PennHIP and OFA measurements of hip joint quality for breeding selection to reduce hip dysplasia in purpose-bred detection dogs. J Am Vet Med Assoc. 2020. PMID 32657662. Link
31. Lewis TW, Blott SC, Woolliams JA. Canine hip dysplasia: phenotypic scoring and the role of estimated breeding value analysis. Vet J. 2014;201:229-236. PMID 25072401. Link
32. Orthopedic Foundation for Animals. Basic OFA Breeding Recommendations. Accessed October 2026. Link
33. Calboli FCF, Sampson J, Fretwell N, Balding DJ. Population structure and inbreeding from pedigree analysis of purebred dogs. Genetics. 2008;179(1):593-601. PMID 18493074. Link

Source and interpretation note

References were selected to emphasize peer-reviewed veterinary and quantitative-genetics literature, supplemented by official OFA methodology and breeding guidance. Some cited studies are observational, some are experimental, and some analyze voluntary registry data; each design has different strengths and limitations. Numerical findings are presented in the context of the studied populations and should not be treated as universal thresholds for every breed or breeding program.